This study investigated the "customization" of a standardized achievement test to measure the effect on group and individual scores. "Customization" refers to the modification of an existing test to more closely match test specifications of the entity using the test, usually a state or large school district. Results of an in-depth study of students in grades 4 and 6 in New York City are reported. Subject sample sizes ranged from 5,200 to 6,700 at grade 4 and from 6,200 to 7,600 at grade 6. Five hypothetically customized tests were constructed at each grade level by selecting five different General Survey Forms (sets of 20 piloted items) and substituting each set into a shelf test--the mathematics test of the Metropolitan Achievement Test, Sixth Edition (MAT6). This was done with the intention of creating the maximum difference between the difficulty of the shelf test and the customized New York test. Analysis of the effect on group and individual scores for the shelf test versus the customized test indicated that the differences occurred in a predictable direction. With the easiest items from the shelf test being removed and relatively difficult items substituted, the derived test was able to discriminate more at the high end of the MAT6 scale. Therefore, greater differences in scores for high scoring individuals were observed. The greatest changes in mean scaled scores were found in the problem solving subtest at both grade levels. (TJH)